As requirements for large-bandwidth services grow increasingly, a larger switching capacity is also required for a switching network. However, due to limitations of technologies for backplanes, energy consumption, and the like, a traditional electric switch cannot meet a requirement for an ever-increasing switching capacity. An optical switch receives widespread attention for its low power consumption, high capacity, and other characteristics.
Currently, an optical switching system usually uses a burst-mode transmission mechanism to transmit an optical signal, that is, there is a spacing between data packets carried in the optical signal. Therefore, a burst signal is generated. FIG. 1A shows a burst signal. Because amplitudes between burst signals differ greatly, when receiving optical signals, an optical receiving system needs to adjust a parameter such as a gain proportion according to a power peak value of each optical signal, so that the optical signals of different strengths are converted into electrical signals of a same strength, thereby ensuring that the optical receiving system can receive the optical signals successfully.
Currently, a common practice of processing a burst signal by an optical receiving system in the industry is: adding a preamble before a data packet carried in each optical signal, to adjust a status of an optical receiver, as shown in FIG. 1B. However, an adjustment process usually requires hundreds of nanoseconds or a few microseconds; as a power value difference between optical data signals and a data transmission rate increase, the adjustment process is prolonged, and a preamble that needs to be added extends correspondingly, causing a disadvantage of resource waste. According to the Ethernet protocol, an Ethernet data frame length ranges from 64 B to 1510 B. In an example of the 10 G Ethernet, a shortest packet requires 50 ns, and if a preamble that requires tens of nanoseconds is added, a waste of about 50% bandwidth is caused.